| Etiological models of complex disease are elusive, as are replicable findings of large effect. Commonly-cited explanations have previously invoked low-frequency genomic variation, allelic heterogeneity at susceptibility loci, variable etiological trajectories, and epistatic effects between multiple loci; these have represented among the most methodologically-challenging issues in molecular genetic studies of complex traits. Major sequencing initiatives, such as the 1,000 Genomes Project, are currently identifying human polymorphic sites at frequencies previously unassailable and, not ten years after publication of the first major genome-wide association findings, medical sequencing has already begun to displace GWAS as the standard for genetic analysis of complex traits. However, several recent studies have shown that the cumulative effect of a large number of common SNPs can account for a significant proportion of the variance in liability to complex traits, highlighting a conspicuous discrepancy between the explanatory value of reported GWAS associations and the realized contribution of common genetic variation. Emergent polygenic models posit the influence of thousands of common causal variants, many or most of which will remain obscured by genome-wide significance thresholds. Expectations regarding the number of additional variants "discoverable" by GWAS are sobering, as are implications for risk prediction in complex disease. With studies of complex disease primed for an unprecedented survey of human genetic variation, it is essential that these nascent, impending challenges be addressed.;Of interest herein are methodologies which utilize differential patterns of linkage disequilibrium to resolve the underlying genetic liability to complex traits, the range of allele frequencies for which common association tests are appropriate, and the relevant dimensionality of common genetic variation within ethnically-concordant but differentially ascertained populations. Using high-density SNP genotype data, we consider both hypothesis-driven and agnostic (genome-wide) approaches to association analysis, and address specific issues pertaining to empirical significance and the statistical properties of commonly-applied tests. Lastly, we attempt to place these diverse contributions into a unified framework of human population genetic theory. |
